Dioxygen Activation in Methane Monooxygenase: A Theoretical Study

Feb 13, 2004 - An oxygen kinetic isotope effect (KIE) of 1.016 (16O16O vs 18O16O), determined for steady-state turnover,50 indicates that the O−O bo...
0 downloads 11 Views 241KB Size
Published on Web 02/13/2004

Dioxygen Activation in Methane Monooxygenase: A Theoretical Study Benjamin F. Gherman,† Mu-Hyun Baik,† Stephen J. Lippard,*,‡ and Richard A. Friesner*,† Contribution from the Department of Chemistry and Center for Biomolecular Simulation, Columbia UniVersity, New York, New York 10027, and Department of Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139 Received June 4, 2003; E-mail: [email protected]

Abstract: Using broken-symmetry unrestricted Density Functional Theory, the mechanism of enzymatic dioxygen activation by the hydroxylase component of soluble methane monooxygenase (MMOH) is determined to atomic detail. After a thorough examination of mechanistic alternatives, an optimal pathway was identified. The diiron(II) state Hred reacts with dioxygen to give a ferromagnetically coupled diiron(II,III) Hsuperoxo structure, which undergoes intersystem crossing to the antiferromagnetic surface and affords Hperoxo, a symmetric diiron(III) unit with a nonplanar µ-η2:η2-O22- binding mode. Homolytic cleavage of the O-O bond yields the catalytically competent intermediate Q, which has a di (µ-oxo)diiron(IV) core. A carboxylate shift involving Glu243 is essential to the formation of the symmetric Hperoxo and Q structures. Both thermodynamic and kinetic features agree well with experimental data, and computed spin-exchange coupling constants are in accord with spectroscopic values. Evidence is presented for pH-independent decay of Hred and Hperoxo. Key electron-transfer steps that occur in the course of generating Q from Hred are also detailed and interpreted. In contrast to prior theoretical studies, a requisite large model has been employed, electron spins and couplings have been treated in a quantitative manner, potential energy surfaces have been extensively explored, and quantitative total energies have been determined along the reaction pathway.

Introduction

Soluble methane monooxygenase (sMMO) is a non-heme, diiron enzyme system found in methanotrophic bacteria that converts dioxygen and methane to alcohol and water.1-3 Over the past years, extensive experimental and theoretical studies on the catalytic hydroxylation mechanism have been conducted, leading to substantial progress in understanding the mechanism at an atomic level of detail.4-7 Although there is general agreement on the sequence of the major steps of the enzymatic reaction, considerable controversy remains concerning the structures and energetics of many of the intermediates and transition states in the catalytic cycle. The cycle consists of four principal species (Figure 1). The oxidized diiron(III) form Hox, for which crystal structure data are available,8-10 is the first † ‡

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Columbia University. Massachusetts Institute of Technology. Hanson, R. S.; Hanson, T. E. Microbiol. ReV. 1996, 60, 439-471. Lipscomb, J. D. Annu. ReV. Microbiol. 1994, 48, 371-399. Liu, K. E.; Lippard, S. J. AdV. Inorg. Chem. 1995, 42, 263-289. Feig, A. L.; Lippard, S. J. Chem. ReV. 1994, 94, 759-805. Wallar, B. J.; Lipscomb, J. D. Chem. ReV. 1996, 96, 2625-2657. Merkx, M.; Kopp, D. A.; Sazinsky, M. H.; Blazyk, J. L.; Mu¨ller, J.; Lippard, S. J. Angew. Chem., Int. Ed. Engl. 2001, 40, 2782-2807. Baik, M.-H.; Newcomb, M.; Friesner, R. A.; Lippard, S. J. Chem. ReV. 2003, 103, 2385-2419. Rosenzweig, A. C.; Frederick, C. A.; Lippard, S. J.; Nordlund, P. Nature 1993, 366, 537-543. Rosenzweig, A. C.; Lippard, S. J. Acc. Chem. Res. 1994, 27, 229-236. Elango, N.; Radhakrishnan, R.; Froland, W. A.; Wallar, B. J.; Earhart, C. A.; Lipscomb, J. D.; Ohlendorf, D. H. Protein Sci. 1997, 6, 556-568.

2978

9

J. AM. CHEM. SOC. 2004, 126, 2978-2990

Figure 1. Overview of the MMOH catalytic cycle.

and represents the resting state of the sMMO hydroxylase (MMOH). Enzymatic catalysis is initiated when dioxygen reacts with the Fe(II) centers of the reduced form of MMOH (Hred) to give Hperoxo, a diiron(III) intermediate.11-14 Hperoxo evolves to intermediate Q,15 a diiron(IV) species,14,16 which reacts with methane to yield methanol and Hox.17-21 (11) Liu, K. E.; Wang, D. L.; Huynh, B. H.; Edmondson, D. E.; Salifoglou, A.; Lippard, S. J. J. Am. Chem. Soc. 1994, 116, 7465-7466. (12) Lee, S. K.; Nesheim, J. C.; Lipscomb, J. D. J. Biol. Chem. 1993, 268, 21 569-21 577. (13) Brazeau, B. J.; Lipscomb, J. D. Biochemistry 2000, 39, 13 503-13 515. (14) Liu, K. E.; Valentine, A. M.; Wang, D. L.; Huynh, B. H.; Edmondson, D. E.; Salifoglou, A.; Lippard, S. J. J. Am. Chem. Soc. 1995, 117, 10 17410 185. (15) Shu, L. J.; Nesheim, J. C.; Kauffmann, K.; Mu¨nck, E.; Lipscomb, J. D.; Que, L. Science 1997, 275, 515-518. (16) Lee, S. K.; Fox, B. G.; Froland, W. A.; Lipscomb, J. D.; Mu¨nck, E. J. Am. Chem. Soc. 1993, 115, 6450-6451. 10.1021/ja036506+ CCC: $27.50 © 2004 American Chemical Society

Dioxygen Activation in Methane Monooxygenase

The present paper is focused on the dioxygen activation steps. We inquire as to how dioxygen is integrated into the diiron core of the MMOH active site to create a catalytically competent species, capable of inserting an oxygen atom into an aliphatic C-H bond at ambient temperature. We22 and others23-32 have addressed this question previously using quantum chemical methods. There is disagreement, however, about the dioxygen activation mechanism and the molecular structure of Hperoxo. A critical comparison between the different proposals has proved difficult owing to fundamentally different strategies and assumptions in constructing the computational models.7 The goal of the present study was therefore to examine in stringent detail the dioxygen activation mechanism using large computational models and quantum chemical methods capable of capturing the electronic structure of the MMOH catalytic species, both necessary (vide infra) features similar to those employed in our previous work.22,33 The results allow us to comment on some controversial experimental findings, outlined in the next section, proposing explanations for the differences where possible. Advances in computing technology, as well as our own experience with the computational models,34 have allowed us to investigate a significantly greater range of structures than were considered previously, in turn enabling a much more rigorous characterization of the dioxygen activation steps. The calculations are particularly demanding, because a large number of spin states exist at various points along the catalytic pathway. We have exhaustively enumerated these states and identified dominant reaction steps based on relative energies, taking into account whether particular transformations are spinallowed. Whereas our previous work considered only intermediates,22 in this study we have also computed transition states, affording rate constants that can be compared with experimental data. These results can be used along with other criteria, such as consistency of computed structures with spectroscopic data for the intermediates and the thermodynamic stability of the various species, to assess the extent to which theory and experiment agree. This process is not simply a matter of judging theory based on its ability to predict experimental results. There are (17) Stirling, D. I.; Dalton, H. Nature 1981, 291, 169. (18) Colby, J.; Stirling, D. I.; Dalton, H. Biochem. J. 1977, 165, 395-402. (19) Fox, B. G.; Borneman, J. G.; Wackett, L. P.; Lipscomb, J. D. Biochemistry 1990, 29, 6419-6427. (20) Andersson, K. K.; Froland, W. A.; Lee, S. K.; Lipscomb, J. D. New J. Chem. 1991, 15, 411-415. (21) Rataj, M. J.; Kauth, J. E.; Donnelly, M. I. J. Biol. Chem. 1991, 266, 18 68418 690. (22) Dunietz, B. D.; Beachy, M. D.; Cao, Y. X.; Whittington, D. A.; Lippard, S. J.; Friesner, R. A. J. Am. Chem. Soc. 2000, 122, 2828-2839. (23) Siegbahn, P. E. M.; Crabtree, R. H.; Nordlund, P. J. Biol. Inorg. Chem. 1998, 3, 314-317. (24) Siegbahn, P. E. M. Inorg. Chem. 1999, 38, 2880-2889. (25) Siegbahn, P. E. M. J. Biol. Inorg. Chem. 2001, 6, 27-45. (26) Yoshizawa, K.; Hoffmann, R. Inorg. Chem. 1996, 35, 2409-2410. (27) Yoshizawa, K.; Yokomichi, Y.; Shiota, Y.; Ohta, T.; Yamabe, T. Chem. Lett. 1997, 587-588. (28) Yoshizawa, K.; Yamabe, T.; Hoffmann, R. New J. Chem. 1997, 21, 151161. (29) Yoshizawa, K.; Ohta, T.; Yamabe, T.; Hoffmann, R. J. Am. Chem. Soc. 1997, 119, 12 311-12 321. (30) Torrent, M.; Musaev, D. G.; Morokuma, K.; Basch, H. J. Phys. Chem. B 2001, 105, 4453-4463. (31) Torrent, M.; Mogi, K.; Basch, H.; Musaev, D. G.; Morokuma, K. J. Phys. Chem. B 2001, 105, 8616-8628. (32) Torrent, M.; Musaev, D. G.; Basch, H.; Morokuma, K. J. Comput. Chem. 2002, 23, 59-76. (33) Gherman, B. F.; Dunietz, B. D.; Whittington, D. A.; Lippard, S. J.; Friesner, R. A. J. Am. Chem. Soc. 2001, 123, 3836-3837. (34) Friesner, R. A.; Baik, M.-H.; Guallar, V.; Gherman, B. F.; Wirstam, M.; Murphy, R. B.; Lippard, S. J. Coord. Chem. ReV. 2003, 238-239, 267290.

ARTICLES

uncertainties and controversies associated with many of the experiments as well, so that the entire process requires assembling a coherent, integrated picture in which both calculations and reported experimental data are critically examined. It is the case, however, that the high quality of the theoretical calculations with regard to system size, basis sets, and Density Functional Theory (DFT)35 methodology, combined with the greatly expanded coverage of phase space with respect to both structural alternatives and spin couplings, enable such a picture to be assembled at a level of detail that has not been possible to date. Although we do not regard the results presented herein to be complete, the level of quantitative theory/experiment comparison has been substantially enhanced, which in turn increases confidence in the conclusions that are drawn. Relevant Experimental Data. An extensive series of experiments has been performed on enzymes isolated from Methylococcus capsulatus (Bath) and Methylosinus trichosporium OB3b, hereafter MMO Bath and MMO OB3b, respectively, to determine the mechanism by which these bacteria oxidize methane. This body of work has established a number of important features that serve as benchmark points for our computational model. The ‘reactant’ in the dioxygen activation process, the reduced form of MMOH (Hred), has EPR and zero-field Mo¨ssbauer spectra indicative of two high-spin iron(II) atoms that are weakly ferromagnetically (F) coupled and in different ligand environments.36-39 X-ray crystallography, circular dichroism (CD) and magnetic circular dichroism (MCD) experiments reveal the iron atoms to be 5-coordinate, with the open coordination sites available for subsequent occupation by dioxygen-derived ligands.40-42 The first step of the catalysis is reaction of Hred with dioxygen. This process is poorly understood at present and many conflicting observations have been made. For example, the formation of Hperoxo was reported to be pH independent in the MMO Bath system,14 whereas a different study using MMO OB3b found a sharp decrease in the Hperoxo formation rate with increasing pH.43 The reported rate constants for the formation of Hperoxo are also different. When measured in the MMO OB3b system, Hperoxo formation had a rate constant of 9-12 s-1 at 4 °C,13 whereas rates of 25 s-1 and more recently 1-2 s-1 have been found using MMO Bath.11,14,44 Similarly puzzling is the observation that the decay rate of Hred is consistently higher than the rate of Hperoxo formation,12,13,44,45 which could mean that there is at least one other species playing a role in the decay of Hred. However, (35) Parr, R. G.; Yang, W. Density Functional Theory of Atoms and Molecules; Oxford University Press: New York, 1989. (36) Dewitt, J. G.; Bentsen, J. G.; Rosenzweig, A. C.; Hedman, B.; Green, J.; Pilkington, S.; Papaefthymiou, G. C.; Dalton, H.; Hodgson, K. O.; Lippard, S. J. J. Am. Chem. Soc. 1991, 113, 9219-9235. (37) Fox, B. G.; Surerus, K. K.; Mu¨nck, E.; Lipscomb, J. D. J. Biol. Chem. 1988, 263, 10 553-10 556. (38) Fox, B. G.; Hendrich, M. P.; Surerus, K. K.; Andersson, K. K.; Froland, W. A.; Lipscomb, J. D.; Mu¨nck, E. J. Am. Chem. Soc. 1993, 115, 36883701. (39) Hendrich, M. P.; Mu¨nck, E.; Fox, B. G.; Lipscomb, J. D. J. Am. Chem. Soc. 1990, 112, 5861-5865. (40) Pulver, S.; Froland, W. A.; Fox, B. G.; Lipscomb, J. D.; Solomon, E. I. J. Am. Chem. Soc. 1993, 115, 12 409-12 422. (41) Rosenzweig, A. C.; Nordlund, P.; Takahara, P. M.; Frederick, C. A.; Lippard, S. J. Chem. Biol. 1995, 2, 409-418. (42) Whittington, D. A.; Lippard, S. J. J. Am. Chem. Soc. 2001, 123, 827-838. (43) Lee, S. K.; Lipscomb, J. D. Biochemistry 1999, 38, 4423-4432. (44) Valentine, A. M.; Stahl, S. S.; Lippard, S. J. J. Am. Chem. Soc. 1999, 121, 3876-3887. (45) Liu, Y.; Nesheim, J. C.; Lee, S. K.; Lipscomb, J. D. J. Biol. Chem. 1995, 270, 24 662-24 665. J. AM. CHEM. SOC.

9

VOL. 126, NO. 9, 2004 2979

ARTICLES

Gherman et al.

no positive identification of such an intermediate has been made to date and its existence is mainly supported by the rate mismatch between Hred decay and Hperoxo formation. There are also a number of features that have been fairly well established. Hperoxo has consistently been the first directly observable intermediate following the decay of Hred.11-14,16 Mo¨ssbauer studies suggest that Hperoxo is diamagnetic owing to antiferromagnetic (AF) coupling of the two iron atoms.11,14 A comparison with spectroscopic data from structurally well-defined Fe(III) peroxide model complexes46-49 point to two electronically equivalent high-spin Fe(III) atoms that are symmetrically bridged by the peroxo moiety. Whether the peroxide group adopts an end-on, a µ-η1:η1, or a µ-η2:η2 coordination geometry could not be resolved, however. An oxygen kinetic isotope effect (KIE) of 1.016 (16O16O vs 18O16O), determined for steady-state turnover,50 indicates that the O-O bond order decreases substantially at the rate-limiting transformation. The last step of the catalytic cycle that we consider in this work is decay of Hperoxo to form Q. Although there is no doubt that the two intermediates are directly connected, the decay rate of Hperoxo matching the formation rate of Q,14,43 there is disagreement in the literature on the actual rate. In the MMO OB3b system, a rate constant kobs of 2.4 s-1 at 4 °C and pH 7 was measured.13 In the same work, it was found that the rate decreases sharply with increasing pH. In a different study using MMO Bath, however, Hperoxo decay had a rate constant of 0.45 s-1 at 4 °C and was independent of pH in the range 6.6 e pH e 8.6.14 Spectroscopic data do concur that Q contains two nearly equivalent high-valent Fe(IV) atoms that are AFcoupled.11,14-16 Computational Methods All calculations were carried out with the Jaguar v4.1 suite of ab initio quantum chemistry programs.51,52 The B3LYP functional53-55 was used for all DFT calculations. To model correctly the open shell orbitals on the high-spin iron atoms in MMOH, an unrestricted DFT (UDFT) methodology was utilized. AF-coupling was achieved through the use of broken symmetry (BS) UDFT wave functions.56 We have discussed the use of this approximation in the context of MMOH modeling in previous work.22 Although the BS/UDFT methodology is not a completely rigorous treatment of the diiron spin manifold, it has proved to be a very useful and practical approximation, providing a substantial increase in realism as compared to simpler approaches, which typically employ F-coupling for all species. The starting model used in this study contains ∼100 atoms of the reduced MMOH active site taken from the coordinates in ref 41 (Figure 2). As compared to smaller models, this large model is critical22 to maintaining key structural features seen in the crystallographic data8,41,57 and therefore in retaining biological relevance. Geometry optimizations (46) Micklitz, W.; Bott, S. G.; Bentsen, J. G.; Lippard, S. J. J. Am. Chem. Soc. 1989, 111, 372-374. (47) Menage, S.; Brennan, B. A.; Juarezgarcia, C.; Munck, E.; Que, L. J. Am. Chem. Soc. 1990, 112, 6423-6425. (48) Brennan, B. A.; Chen, Q. H.; Juarezgarcia, C.; True, A. E.; Oconnor, C. J.; Que, L. Inorg. Chem. 1991, 30, 1937-1943. (49) Kim, K.; Lippard, S. J. J. Am. Chem. Soc. 1996, 118, 4914-4915. (50) Stahl, S. S.; Francisco, W. A.; Merkx, M.; Klinman, J. P.; Lippard, S. J. J. Biol. Chem. 2001, 276, 4549-4553. (51) Jaguar v4.1, Schro¨dinger, Inc., Portland, Oregon, 2000. (52) Friesner, R. A.; Murphy, R. B.; Beachy, M. D.; Ringnalda, M. N.; Pollard, W. T.; Dunietz, B. D.; Cao, Y. X. J. Phys. Chem. A 1999, 103, 19131928. (53) Johnson, B. G.; Gill, P. M. W.; Pople, J. A. J. Chem. Phys. 1993, 98, 5612-5626. (54) Becke, A. D. J. Chem. Phys. 1993, 98, 1372-1377. (55) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. ReV. B 1988, 37, 785-789. (56) Noodleman, L. J. Chem. Phys. 1981, 74, 5737-5743. 2980 J. AM. CHEM. SOC.

9

VOL. 126, NO. 9, 2004

Figure 2. Reduced MMOH crystal structure, adapted from ref 41. Only residues coordinated to the iron atoms are labeled. Hydrogen atoms on the protein residues are omitted for clarity.

were carried out with the 6-31G basis set, except for the inclusion of polarization functions (i.e., the 6-31G** basis set) on selected atoms near the dinuclear iron core. The Los Alamos LACVP** basis set,58-60 with an effective core potential and polarization functions, was used for the two iron atoms. To obtain accurate relative energies of qualitatively different structures, single-point calculations were performed on the optimized geometries. For atoms near the diiron core, the cc-pVTZ(-f) basis set of Dunning61 was used in these calculations, whereas the 6-31G** basis was applied to atoms on the periphery of the model. The iron atoms were treated with the LACV3P** basis set, a triple-ζ quality basis compatible with the Fe effective core potential.58-60 Mappings of energy surfaces, performed along every reaction coordinate in the study, were achieved by freezing one or more coordinates at different values and optimizing all other degrees of freedom. All mappings were performed with the small mixed basis set. Use of the large mixed basis set in this case significantly increases computational time, but only minimally affects relative energies along the reaction coordinates.33 Energetic maxima along the reaction coordinates were then used as starting structures for transition state searches, which were carried out using the quadratic synchronous transit (QST) method.62 In this approach, the initial part of the transition state search is restricted to a circular curve connecting the reactant, transition state guess, and product, after which the optimizer follows the Hessian eigenvector most similar to the tangent of the circular curve. Vibrational frequencies were calculated analytically. Due to the expense of these calculations, the full MMOH model (Figure 2) was truncated to approximately 40-45 atoms (Figure 3), which decreased the cost of the frequency calculations by roughly 90%. Glutamate residues coordinated to the iron atoms were cut between CR and Cβ, and capped with hydrogen atoms. Histidine residues were replaced by NH3 ligands. The positions of the capping carboxylate and ammonia hydrogen atoms were then optimized with the structure otherwise frozen, and the results were used for the frequency calculation. From the frequency calculations, zero point energy corrections as well as enthalpy and entropy corrections can be obtained, thereby allowing free energy differences at room temperature to be calculated for all reactions. Mulliken spin density analysis has become a routine tool to confirm that a reasonable electronic structure has been reached.7,34 In addition, it provides a convenient method of tracing electron flow. By definition, (57) Rosenzweig, A. C.; Brandstetter, H.; Whittington, D. A.; Nordlund, P.; Lippard, S. J.; Frederick, C. A. Proteins: Struct., Funct., Genet. 1997, 29, 141-152. (58) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 270-283. (59) Wadt, W. R.; Hay, P. J. J. Chem. Phys. 1985, 82, 284-298. (60) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299-310. (61) Dunning, T. H. J. Chem. Phys. 1989, 90, 1007-1023. (62) Peng, C. Y.; Schlegel, H. B. Isr. J. Chem. 1993, 33, 449-454.

Dioxygen Activation in Methane Monooxygenase

Figure 3. Truncated Hred model. Only residues coordinated to the iron atoms are labeled. Hydrogen atoms on the protein residues are omitted for clarity.

R-electrons are assigned positive spin and β-electrons negative spin. In principle, one would expect a spin density of 4.0 for each Fe(II)center in Hred owing to the four unpaired electrons in a high-spin d6 configuration. Due to electronic metal-ligand interactions and the intrinsic formalism of assigning electrons to a specific atom, however, noninteger spin densities lower than the ideal number are commonly observed. A spin density of 3.5-3.7, for example, is a typical signature for a high-spin Fe(II)-d4 center, whereas a spin density of 3.8-4.1 indicates an Fe(III)-d5 center.

Results

A. Reduced Enzyme Hred and Preparation for Attack by Dioxygen. The diiron(II) state was modeled with high-spin S ) 2, F-coupled iron atoms in accord with experimental evidence,14,36-39 giving a 9A state, Hred-F. The case of AFcoupling, i.e., the 1A state Hred-AF, is discussed in the Supporting Information (text section A). As in the crystal structure,41,42 geometry optimization maintained a loosely coordinated water molecule, trans to the histidine coordinated to Fe1. The optimized model had an Fe-Fe distance about 0.35 Å longer than in the crystal structure, reflecting the energetically flat FeFe potential surface, an issue discussed previously in ref 22. The discrepancy in distance may reflect relatively subtle errors in the DFT functional, but has a minimal effect on the major structural and energetic features of the reaction chemistry discussed below. In fact, the energy change as a function of this coordinate, over the relevant range of distances, is extremely small, on the order of 1-2 kcal/mol at most. The first step in the reaction of Hred with O2 involves departure of the loosely coordinated water molecule trans to the histidines. Figure S2 depicts the reaction coordinate for this process, which is energetically uphill while the water molecule remains in the vicinity of the diiron core. As the water moves away from the core, increasing entropy lowers the system free energy. Computation of the final free energy change upon departure of the water involves the summation of various energy components. First, if the water molecule is considered to be released into the gas phase, ∆H for Hred f Hred-primed + H2O(g) is +16.39 kcal/mol, including zero-point energy and enthalpy corrections at 25 °C. In reality, however, the water molecule is released into solution, and the gas phase-to-solution free energy of transfer for this process must also be included. The enthalpy change for Hred f Hred-primed + H2O(aq) then becomes +8.92 kcal/mol.63,64 Second, when the water molecule departs the

ARTICLES

protein environment for solution, van der Waals interactions with the protein are lost at a cost of approximately 6 kcal/mol.65 This correction is necessary because DFT methods do not account for vdW interactions,66,67 which would lead to a serious error for the energy of the water-containing species (i.e., Hred). The final enthalpy change is then +14.92 kcal/mol.64 The entropic correction (the translational/rotational entropy gain) for release of the water molecule is significant, 13.09 kcal/mol. The final free energy change is therefore +1.83 kcal/mol, and the reaction is thus nearly isoenergetic. The spin and charge distributions (Table S10) are only slightly different from those seen for Hred. The Hred-primed structure created by the loss of the water molecule has open coordination sites on Fe1 and Fe2, each being 5-coordinate, which are available for dioxygen binding in the next steps of the reaction. B. Attack of Dioxygen and Formation of the Peroxo Intermediate Hperoxo. B.1. Overview. The next step in reductive activation of dioxygen is for O2 to enter the enzyme active site, bind to the diiron core, and form Hperoxo. There has been some controversy as to whether there are stable intermediates prior to Hperoxo. Our calculations address this question in what follows. The attack of dioxygen upon the diiron core is complicated by several important issues: (1) One or two electrons can be transferred from the diiron core to dioxygen, forming superoxo and peroxo moieties, respectively. Furthermore, these transfers can take place at different points along the reaction coordinate. (2) There are a number of different possible Hsuperoxo or Hperoxo structures. (3) There are a large number of possible ways to couple the spins on the oxygen atoms of dioxygen with the spins on the two iron atoms. We have addressed these three issues by modeling the various relevant alternatives, comparing the activation free energies of the possible kinetic pathways as well as assessing whether they would be spin-forbidden, which would diminish the effective rate constant substantially. The optimal pathway is then selected based on energetic and spin compatibility considerations (Supporting Information text sections C and D). Here, we provide a brief overview of the manifold of structures and electronic states, and then focus on characterization of the optimal reaction pathway from both a qualitative and quantitative point of view. B.2. Hsuperoxo and Hperoxo Structures and Spin States. We begin our analysis by enumerating the Hsuperoxo and Hperoxo structural/spin intermediates. These were generated by a series of computations employing different starting points, which in turn were based on the experimental data and physical/chemical intuition. All of these structures represent minima on the potential energy surface. Three different Hsuperoxo geometries, the cores of which are shown in Figure 4, were considered to be possibilities. Hsuperoxo-(a) was previously explored by us22 and differs from (63) The solvation energy of water has been computed using a continuum solvation model and amounts to 7.47 kcal/mol. (64) The energetic corrections for solvation used here both for O2(aq) and H2O(aq) are solvation free energies. Adding these free energies of solvation to the computed enthalpies for the model systems is technically not appropriate and then creates a ‘hybrid-energy’ that is not ∆H in a rigorous sense. However, it is impossible to separate the free energy of solvation obtained from Henry’s law or the continuum solvation model into an enthalpic and entropic term. For simplicity, we neglect this detail and use ∆H to denote the solvation-corrected enthalpy. (65) Wirstam, M.; Lippard, S. J.; Friesner, R. A. J. Am. Chem. Soc. 2003, 125, 3980-3987. (66) Kristya´n, S.; Pulay, P. Chem. Phys. Lett. 1994, 229, 175-180. (67) Pe´rez-Jorda´, J. M.; Becke, A. D. Chem. Phys. Lett. 1995, 233, 134-137. J. AM. CHEM. SOC.

9

VOL. 126, NO. 9, 2004 2981

ARTICLES

Gherman et al.

Figure 4. Three general possible geometries for Hsuperoxo. Only cores of the structures are shown. Dashed lines represent hydrogen bonds.

Figure 5. Electron (d electrons for Fe1 and Fe2; π* electrons for dioxygen) spins and couplings for the reaction of Hred-primed-F + O2 f Hsuperoxo-F.

Hsuperoxo-(b) and Hsuperoxo-(c) by the positioning of Glu243. In each structure, stabilization of the O3-O4 superoxo ligand is provided either by a hydrogen bond to O3 (a and b) or by an Fe2-O3 bond accompanied by a weak O3-water interaction (c). In principle, each of the above Hsuperoxo structures can exist in a variety of electronic states. The two Fe(II) centers in Hredprimed-F give rise to a spin multiplicity of 9, with the 8 unpaired parallel spins taken to be spin up. As the triplet dioxygen approaches Hred-primed-F, its two unpaired electrons may both be aligned parallel or antiparallel to the electrons in Hredprimed-F (Figure 5). If the spins of the dioxygen-based electrons are parallel to those on the iron atoms, the total multiplicity of the system is 11. Oxidation of one of the iron atoms yields a Hsuperoxo structure, which still has multiplicity of 11. We have determined that Fe2 is preferentially oxidized in Hsuperoxo motifs (b) and (c) for all spin states considered (Tables S11, S12). If the two unpaired electrons on dioxygen are initially antiparallel to the spins in Hred-primed-F, then the net multiplicity of the combined system is 7. After Fe2 is oxidized and dioxygen reduced, the resulting Hsuperoxo will have multiplicity 9 as illustrated in Figure 5. With reactants on the multiplicity 7 surface and products on the multiplicity 9 surface, intersystem crossing (ISC) must occur at some point along the reaction coordinate. In the AF-coupled case (Figure S3), a triplet or singlet Hsuperoxo structure can be generated from the singlet Hredprimed-AF state in an analogous fashion to that for the multiplicity 11 and 9 F-coupled Hsuperoxo, respectively (Supporting Information text section B). Core geometries of the two possible Hperoxo species are presented in Figure 6. The asymmetric Hperoxo structure was first presented elsewhere.22 In the symmetric structure, Glu243 shifts so that O6 is hydrogen bonded to the water molecule coordinated to Fe1 rather than bridging the two iron atoms. The symmetric 2982 J. AM. CHEM. SOC.

9

VOL. 126, NO. 9, 2004

Figure 6. Two general possible geometries for Hperoxo. Only cores of the structures are shown. Dashed lines represent hydrogen bonds.

structure also has the peroxide moiety in a symmetric nonplanar bridging µ-η2:η2 butterfly arrangement, similar to that which occurs following dioxygen activation by some dicopper complexes.68-78 In each structure, stabilization is provided by a key hydrogen bond, whether it be O3-water in the asymmetric geometry or O6-water in the symmetric structure. The spin coupling alternatives for Hperoxo are considerably simpler than those for Hsuperoxo. The only unpaired electrons in the Hperoxo intermediate are the 5 d-electrons on each Fe(III) atom, which can F-couple to give a multiplicity of 11 or AF-couple to give a multiplicity of 1. (68) Kitajima, N.; Morooka, Y. J. Chem. Soc., Dalton Trans. 1993, 26652671. (69) Kitajima, N.; Morooka, Y. Chem. ReV. 1994, 94, 737-757. (70) Morooka, Y.; Fujisawa, K.; Kitajima, N. Pure Appl. Chem. 1995, 67, 241248. (71) Karlin, K. D.; Kaderli, S.; Zuberbuhler, A. D. Acc. Chem. Res. 1997, 30, 139-147. (72) Liang, H. C.; Dahan, M.; Karlin, K. D. Curr. Opin. Chem. Biol. 1999, 3, 168-175. (73) Siegbahn, P. E. M.; Wirstam, M. J. Am. Chem. Soc. 2001, 123, 11 81911 820. (74) Solomon, E. I.; Chen, P.; Metz, M.; Lee, S. K.; Palmer, A. E. Angew. Chem. Int. Ed. Engl. 2001, 40, 4570-4590. (75) Que, L.; Tolman, W. B. Angew. Chem., Int. Ed. Engl. 2002, 41, 11141137. (76) Metz, M.; Solomon, E. I. J. Am. Chem. Soc. 2001, 123, 4938-4950. (77) Blackman, A. G.; Tolman, W. B. Struct. Bond. 2000, 97, 179-211. (78) Mahadevan, V.; Gebbink, R. J. M. K.; Stack, T. D. P. Curr. Opin. Chem. Biol. 2000, 4, 228-234.

Dioxygen Activation in Methane Monooxygenase

ARTICLES

Figure 7. Schematic reaction energy profile for reductive dioxygen activation by MMOH. Table 1. Relative Enthalpies, Entropies, and Free Energies (all at 25 °C) for the Alternative Hsuperoxo and Hperoxo Structures species

∆H (kcal/mol)a

∆S (cal/mol*K)

∆G (kcal/mol)a

Hred-primed-F + O2 (aq) 9A H superoxo (9SUP) 11A H superoxo (11SUP) 1A H superoxo (1SUP) 3A H superoxo (3SUP) 11A asymmetric H peroxo (11APER) 11A symmetric H peroxo (11SPER) 1A asymmetric H peroxo (1APER) 1A symmetric H peroxo (1SPER)

0.00 +0.27 -4.10 -5.61 -4.29 -3.57 -10.72 -4.32 -12.29

0.00 -42.45 -34.43 -42.90 -34.31 -45.08 -42.80 -50.21 -46.26

0.00 +12.93 +6.17 +7.18 +5.94 +9.88 +2.04 +9.04 -0.11

a Enthalpies and free energies include contributions from the solvation energy of O2 and from the van der Waals interaction of O2 with Hred-primed-F.

Table 1 summarizes these alternative intermediate structures and their total energies relative to Hred-primed-F. A number conclusions can be drawn immediately: (1) The lowest energy intermediate is the symmetric peroxo species 1SPER (cf., Figure 6b), which is the only one that is comparable in energy to the starting complex. It will be expected to have significant population and hence be spectroscopically observable. We therefore identify species 1SPER with Hperoxo. The suitability of species 1SPER as a candidate for Hperoxo is further discussed below. (2) The absence of other low energy intermediates is consistent with the inability to detect any such species directly despite extensive experimental investigation of MMOH over the past decade. Some indirect evidence has led to speculation about the existence of other intermediates; we consider these arguments briefly in section A.1. of the Discussion section. Our calculations suggest that alternative explanations of these experiments are to be preferred. (3) Identification of species 1SPER with Hperoxo leaves open the question of the optimal kinetic pathway to this state from the initial complex. Clearly there are a large number of alternatives, considering that one could pass through any of the states enumerated in Table 1. We have exhaustively investigated the various possible transition states for each putative process, and these results are presented in detail in the Supporting Information (text section C). The

Table 2. Relative Enthalpies, Entropies, and Free Energies (all at 25 °C) for the Species in Figure 8a species

∆H (kcal/mol)b

∆S (cal/mol*K)

∆G (kcal/mol)b

Hred-F + O2 (aq) Hred-primed-F + H2O (aq) + O2 (aq) 11TST-11SUP 11SUP 3SUP 1SUP 1TST-1SPER 1SPER 1TST-1Q 1MV-1Q 1Qc

0.00 +14.92

0.00 +43.92

0.00 +1.83

+13.35 +10.82 +10.63 +9.31 +22.41 +2.68 +19.75 +11.14 -7.93

+7.78 +9.49 +9.61 +1.02 +2.32 +3.06 +0.10 -5.56 +0.44

+11.04 +8.00 +7.84 +9.08 +21.79 +1.79 +19.74 +12.82 -8.04

a All intermediates and transition states from 11TST-11SUP onward include the energy for H2O (aq). b Enthalpies and free energies include contributions from the solvation energy of O2 and from the van der Waals interaction of O2 with Hred-primed-F. c Free energy does not include singlet stabilization energy arising from spin contamination considerations (see results section E).

optimal pathway, extracted from this analysis, is discussed in the next section. B.3. Optimal Kinetic Pathway for the Formation of Hperoxo. The optimal kinetic pathway for the formation of Hperoxo, taking into account energetics and spin compatibility (Supporting Information text section D), is depicted in Figure 7. Relative energies for the states indicated in the Figure are listed in Table 2, and Mulliken spin densities are listed in Table 3. The structures labeled in Figure 7 are depicted with selected geometric information in Figure 8. The rate determining transition state is the species 1TST-1SPER, in which the Hsuperoxo species 1SUP converts to Hperoxo. The reaction coordinate for formation of the first intermediate in dioxygen activation, 11A Hsuperoxo (11SUP), is shown in Figure S7. Energy mapping for dioxygen approach along the 11A surface reveals that the transition state (11TST-11SUP) occurs when O4 is 2.30-2.50 Å from the midpoint between the two iron atoms. The 11A Hsuperoxo that forms on this surface corresponds to a shallow minimum at an O4-diiron midpoint distance of ∼1.90 Å. To calculate free energies for these species, J. AM. CHEM. SOC.

9

VOL. 126, NO. 9, 2004 2983

ARTICLES

Gherman et al.

Table 3. Mulliken Spin Densities for the Species in Figure 8 species

Fe1

Fe2

O3

O4

Hred-F Hred-primed-F 11TST-11SUP 11SUP 3SUP 1SUP 1TST-1SPER 1SPER 1TST-1Q 1MV-1Q 1Q

3.77 3.75 3.76 3.70 -3.76 -3.75 -3.92 -4.02 -3.73 -3.13 -3.49

3.75 3.73 3.80 3.89 3.92 4.04 4.00 4.00 4.01 3.86 3.54

n/a n/a 0.85 0.78 0.75 -0.60 -0.09 -0.10 -0.25 -0.53 -0.06

n/a n/a 0.92 0.79 0.75 -0.23 -0.29 -0.09 -0.35 -0.67 -0.03

we estimate the van der Waals energy for dioxygen within the MMOH substrate binding cavity to be ∼6 kcal/mol,65 an essential correction to the DFT energies of the dioxygen adducts. Second, since the final energy should reflect the equilibrium between dioxygen bound to the protein and O2(aq), we account for the solubility of O2 in water, which, from Henry’s Law, is -3.95 kcal/mol at 25 °C. The transition state 11TST-11SUP then has an enthalpy of 13.35 kcal/mol using the relative energy scale shown in Figure 7 and the entropy correction amounts to 7.78 eu with respect to Hred-F and free dioxygen. It is interesting to note that the exchange of the water ligand with the dioxygen moiety in its superoxo form is entropically favorable, while being enthalpically uphill. Overall the free energy of activation is 11.04 kcal/mol. The transition state geometry 11TST-11SUP (Figure 8) has a slightly elongated (0.1 Å) Fe-Fe distance, a lengthened O3O4 bond compared to 1.215 Å in dioxygen, and an Fe2-O3 bond that has formed to a notable extent at the transition state. Formation of an Fe2-O4 bond leads to Hsuperoxo (11SUP), in which the O3-O4 bond has lengthened further. In both species, interaction with the water ligand bound to Fe1 and the bidentate binding of the superoxide moiety to Fe2 lead to energetic stabilization and indicate an advantage of side-on attack by dioxygen. End-on attack yields a structure with an unstable O3 atom directed outward from the diiron core. Crystal structures for Hox,8-10,41,42 however, indicate there is water present in the substrate binding cavity which could possibly remain in the Hsuperoxo intermediate and stabilize the superoxo moiety resulting from end-on dioxygen attack. Inclusion of such water molecules into the theoretical model will be undertaken in future QM/ MM calculations. ISC from the F to AF surface, facilitated by near energetic degeneracy as well as significant structural similarity, occurs next between 11SUP and the 3A Hsuperoxo structure 3SUP. A second ISC to the energetically close 1A Hsuperoxo (1SUP) follows, corresponding to a spin flip of the one unpaired electron formally assigned to the superoxo moiety (Figure S3). 1SUP has an Fe1-O4 bond, a shortened Fe2-O4 bond, and a decreased iron-iron distance. 1SUP is stabilized by a hydrogen bond from O3 to the water ligand. The free energy of activation versus 1SUP to form the AFcoupled symmetric Hperoxo structure 1SPER is +12.71 kcal/ mol passing through the transition state 1TST-1SPER, which has a relative free energy of +21.79 kcal/mol. The thermodynamic driving force ∆G for this reaction is -7.29 kcal/mol versus 1SUP (Table 2) placing the Hperoxo species at a free energy of 1.79 kcal/mol on the relative energy scale shown in Figure 7. The reaction involves an initial shifting of Glu243 so 2984 J. AM. CHEM. SOC.

9

VOL. 126, NO. 9, 2004

that O6 shifts from its bridging position between the two iron atoms. Glu243 then continues to move toward its final position in 1SPER, where O6 forms a hydrogen bond to the water ligand on Fe1. Simultaneously, O3 rotates around the Fe2-O4 bond axis so as to occupy the bridging position formerly held by O6. The transition state (1TST-1SPER) reflects a geometry in which the Glu243 carboxylate shift and the rotation of O3 are both partially complete. This transition state is stabilized by a persistent weak interaction between O3 and the water ligand and by the early formation of an Fe2-O3 bond; the Fe1-O3 bond does not form until the reaction is nearly complete. The O3-O4 bond length is minimally changed at the transition state but becomes elongated by 0.15 Å to 1.478 Å, a typical peroxo distance, with formation of the µ-η2:η2 butterfly geometry. C. Formation of Q from Hperoxo. The decay of 1SPER to form intermediate Q, a di-(µ-oxo)diiron(IV) species, is facilitated by cleavage of the O3-O4 bond and contraction of the Fe-Fe distance. In the process, the O22- peroxo moiety is formally reduced to two O2- ions and each iron(III) is oxidized to iron(IV). The proposed mechanism involves homolytic cleavage of the oxygen-oxygen bond. The 1A transition state 1TST1Q, ∆G‡ ) +17.95 kcal/mol versus 1SPER (Table 2), for this reaction (Figure 8) has already cleaved the O3-O4 bond by increasing this distance by 0.42 Å compared to that in 1SPER. This process is accompanied by a 0.33 Å decrease in the FeFe distance. Following this transition state, an unstable 1A mixed-valent Fe(IV)Fe(III) state (1MV-1Q) can be identified with a free energy 6.92 kcal/mol lower than 1TST-1Q. The Fe-Fe distance is decreased by another 0.32 Å and the O3O4 distance is increased by another 0.35 Å in 1MV-1Q. A related mixed-valent intermediate has been produced previously by cryoscopic reduction of intermediate Q and observed with Mo¨ssbauer spectroscopy.79 With the oxidation of Fe1 complete, at this point the Fe1-O3 and Fe1-O4 bond lengths are ∼0.30 Å shorter than those for Fe2-O3 and Fe2-O4. The mixedvalent intermediate, with ∆G ) +11.03 kcal/mol versus 1SPER, decays in an entirely downhill reaction to give intermediate Q (1Q). In this last step, the Fe-Fe distance shortens by 0.26 Å, the O3-O4 distance lengthens by 0.12 Å, and the Fe2-bridging oxygen distances decrease by about 0.2 Å. The 1Q thus formed has both iron atoms in nearly equivalent coordination environments. Throughout the 1SPER f 1Q reaction, the hydrogen bond between O6 and the water ligand shortens and strengthens. Formation of 1Q from 1SPER is 9.83 kcal/mol exothermic. The necessity of generating the catalytically active di-(µ-oxo) core of Q indicates the importance of the carboxylate shift that occurs between Hsuperoxo and 1SPER. Without the Glu243 O6 atom repositioning to form the hydrogen bond, O3 would not have been able to serve as the second bridging oxygen atom in either 1SPER or 1Q. Precedence for the lability of the carboxylate groups in the MMOH active site and related units has been established experimentally42,80-83 and theoretically.84,85 (79) Valentine, A. M.; Tavares, P.; Pereira, A. S.; Davydov, R.; Krebs, C.; Hoffman, B. M.; Edmondson, D. E.; Huynh, B. H.; Lippard, S. J. J. Am. Chem. Soc. 1998, 120, 2190-2191. (80) Nordlund, P.; Eklund, H. Curr. Opin. Struct. Biol. 1995, 5, 758-766. (81) Rardin, R. L.; Tolman, W. B.; Lippard, S. J. New J. Chem. 1991, 15, 417430. (82) Tolman, W. B.; Liu, S. C.; Bentsen, J. G.; Lippard, S. J. J. Am. Chem. Soc. 1991, 113, 152-164. (83) Lee, D.; Lippard, S. J. Inorg. Chem. 2002, 41, 2704-2719. (84) Torrent, M.; Musaev, D. G.; Morokuma, K. J. Phys. Chem. B 2001, 105, 322-327.

Dioxygen Activation in Methane Monooxygenase

ARTICLES

Figure 8. Detail of the proposed reaction pathway for dioxygen activation in MMOH. Cores of the minimized structures are shown. Single numbers indicate distances (Å). Dashed lines represent hydrogen bonds.

In the Fe(IV)Fe(III) mixed-valent structure, the singlet peroxide moiety originating from dioxygen has been reduced by one electron, yielding two bridging oxygen atoms which together carry a formal charge of -3 and one unpaired spin, which may be either spin up or spin down (Figure S18). The spin down case leads to an overall multiplicity of one and was just discussed in detail above. The spin up case leads to a multiplicity of 3, prompting the exploration of the triplet counterparts of 1TST-1Q and 1MV-1Q. These 3A counterparts (Figure S19, Table S13) were structurally similar, but had electronic energies 24.66 and 6.21 kcal/mol higher than 1TST1Q and 1MV-1Q, respectively. This result placed the 3A surface significantly above the 1A surface for the 1SPER f 1Q reaction and ruled out the participation of 3A transition states and intermediates in this reaction. D. Comparison of Computed Free Energy Barriers with Experimental Rate Constants. From the free energy barriers determined along the optimal pathway, rate constants for each (85) Baik, M. H.; Lee, D.; Friesner, R. A.; Lippard, S. J. Isr. J. Chem. 2001, 41, 173-186.

segment of dioxygen activation can be readily computed from transition state theory (eq 1)

kobs ) ( kBTκ/h ) exp (-∆G‡/RT)

(1)

where kobs is the observed rate constant, ∆G‡ is the activation barrier, kB is Boltzmann’s constant, T is temperature, κ is the transmission coefficient (taken to be approximately 1), h is Planck’s constant, and R is the gas constant. Table 4 presents the free energy barriers for the formation of each key intermediate and the associated rate constants. Conversely, experimentally determined rate constants may be translated into free energies of activation (Table 4) according to eq 2. These results are discussed further in the Discussion section below.

∆G‡ ) -RT ln (kobsh/kBTκ)

(2)

E. Calculation of Spin-Exchange Coupling Constants. The present results can be used to obtain theoretical spin-exchange coupling constants for each intermediate involved in dioxygen J. AM. CHEM. SOC.

9

VOL. 126, NO. 9, 2004 2985

ARTICLES

Gherman et al.

Table 4. Theoretical and Experimental Free Energies of Activation and Rate Constants for Formation of Intermediates along the Proposed Pathway for Dioxygen Activation in MMOH theoretical ‡

total ∆G at 4 °C (kcal/mol)

species formed

Hsuperoxo (11SUP)

11.44

experimental ‡

rate constant at 4 °C (sec-1)

total ∆G at 4 °C (kcal/mol)

5.50 × 103

symmetric Hperoxo (1SPER)

22.09

2.20 × 10-5

intermediate Q (1Q)

17.86

4.77 × 10-2

n/a with MMOB 15.80-16.18 14.82-14.97 14.41 with MMOB 16.62 15.70

Table 5. Energy Differences, Coupling Constants, and Singlet Stabilization Energies for HS versus BS States for Key Intermediates Involved in Dioxygen Activation in MMOH

species

Hred Hred-primed Hsuperoxo (3A, 11A) Hsuperoxo (1A, 9A) asymmetric Hperoxo symmetric Hperoxo Q

electronic energy J (cm-1) difference: AF − F experimental (kcal/mol) computed value

+0.13 -2.13 -0.13 -8.22 -0.69 -3.95 -10.23

+2.8 +0.3539 -46.6 n/a -2.3 n/a -143.7 n/a -9.7 n/a -56.0 n/aa -223.6